Background: Thyroid cancer (TC) is the most common endocrine malignancy worldwide. The incidence of TC is high and increasing worldwide due to continuous improvements in diagnostic technology. TC is still often overtreated due to a lack of reliable diagnostic biomarkers. Therefore, determining accurate prognostic predictions to stratify TC patients is important.Methods: Raw data were downloaded from the TCGA database, and pairwise comparisons were applied to identify differentially expressed immune-related lncRNA (DEirlncRNA) pairs. Then, we used univariate Cox regression analysis and a modified Lasso algorithm on these pairs to construct a risk assessment model for TC. Next, TC patients were assigned to high- and low-risk groups based on the optimal cutoff score of the model for the 1-year ROC curve. We evaluated the signature in terms of prognostic independence, predictive value, immune cell infiltration, ICI-related molecules and small-molecule inhibitor efficacy. Results: We identified 30 DEirlncRNA pairs through Lasso regression, and 14 pairs served as the novel predictive signature. The high-risk group had a significantly poorer prognosis than the low-risk group. Cox regression analysis revealed that this immune-related signature can predict prognosis independently and reliably for TC. With the CIBERSORT algorithm, we found an association between the signature and immune cell infiltration. Additionally, several immune checkpoint inhibitor (ICI)-related molecules, such as PD-1 and PD-L1, showed a negative correlation with the high-risk group. We further found that some commonly used small-molecule inhibitors, such as sunitinib, were related to this new signature. Conclusions: We constructed a prognostic immune-related lncRNA signature that can predict TC patient survival without considering the technical bias of different platforms, and this signature also sheds light on TC overall prognosis and novel clinical treatments, such as ICB therapy and small molecular inhibitors.